Correspondence: Xuesong Li and Liqun Zhou, Department of Urology, Peking University First Hospital, Institute of Urology, Peking University, National Urological Cancer Center, No. 8 Xishiku St, Xicheng District, Beijing 100034, China.
To analyse the predictive factors for worse pathological outcome (muscle invasive pT2+, non-organ-confined pT3+ or N+ and histological Grade 3) of upper tract urothelial carcinoma (UTUC) in a Chinese population from a nationwide high-volume centre in China.
Patients and Methods
Predictors were studied by retrospectively reviewing the clinicopathological data of 729 consecutive patients with UTUC treated in our centre from January 2002 to December 2010.
Univariate and multivariate logistic regression analyses were used.
There were more female patients (56.4%) than males and more tumours were located in the ureter (52.7%) than in the pelvis.
In multivariate analysis, male gender (hazard ratio [HR] 1.898, P = 0.001), sessile architecture (HR 3.249, P < 0.001), high grade (HR 5.007, P < 0.001), ipsilateral hydronephrosis (HR 4.768, P < 0.001), renal pelvis location (HR 2.620, P < 0.001) and tumour without multifocality (HR 1.639, P = 0.028) were predictive factors for muscle-invasive UTUC.
Male gender (HR 2.132, P < 0.001), renal pelvis location (HR 3.466, P < 0.001), tumour without multifocality (HR 2.532, P = 0.001), sessile tumour architecture (HR 3.274, P < 0.001), and high grade (HR 3.019, P < 0.001) were predictive factors for non-organ-confined disease.
Chronological old age (HR 1.047, P < 0.001), sessile tumour architecture (HR 25.192, P < 0.001), ipsilateral hydronephrosis (HR 1.689, P = 0.024), and positive urinary cytology (HR 1.997, P = 0.006) were predictive factors for histological Grade 3 UTUC.
There was a predominance of female patients and ureteric tumours in UTUCs of this Chinese population.
Male gender, sessile architecture, tumour location, tumour without multifocality, high histological grade and preoperative ipsilateral hydronephrosis were independent predictive factors for worse pathological outcome of UTUCs.
Radical nephroureterectomy (RNU) is the standard treatment for upper urinary tract urothelial carcinoma (UTUC) . However, conservative treatment strategies, which include endoscopic tumour ablation and segmental ureteric resection, are reasonable for UTUCs of low grade and stage. Conservative management also has the benefit of kidney preservation and sparing the patient the morbidity associated with open radical surgery [2-4]. High tumour stage, high grade and lymph node metastasis (LNM) are significantly associated with poor prognosis [4-6]. For patients with advanced disease, clinical experience gained from management of bladder UC supports the therapeutic value of intraoperative lymphadenectomy and adjuvant chemotherapy. Several reports also support the curative value of lymphadenectomy in muscle-invasive disease (MID, pT2–4) [6-9]. Based on clinical experience with the treatment of muscle-invasive bladder UC, platinum-based adjuvant chemotherapy may improve the outcomes of patients with non-organ-confined disease (NOC, pT3–4 and/or N+). However, the loss of renal function after RNU may render a patient with UTUC ineligible for treatment with platinum after surgery. Neoadjuvant chemotherapy is another management option [10, 11]. Ultimately, the accurate prediction of worse pathological outcome, e.g. MID, NOC and Grade 3 UTUCs, may assist the urologist in clinical risk stratification and devising therapeutic options.
The incidence of UTUCs in the Chinese population is quite different from Western populations. In Western countries, UTUCs account for only 5–10% of TCCs and ≈10% of renal tumours . Ureteric TCC is less common than renal pelvis TCC, and accounts for only 30–40% of all UTUCs [4, 12]. UTUCs are two- or three-times more prevalent in men than in women . And female patients have a higher proportion of UTUCs with higher stage, higher grade and with LNM than males [13, 14]. Female patients with UTUCs also had a higher proportion of renal pelvis TCC than male counterparts [4, 12].
However, in the Chinese population we have observed very different patterns related to incidence and disease presentation. First, UTUCs are more common in women instead of men: the male to female ratio is about 1 to 1.3 [15, 16]. Second, UTUCs are more common than in Western populations: they account for 20–30% of all TCCs [16, 17]. Third, renal pelvis TCC is over twice as common among Chinese than among Western populations. Ureter TCC accounts for more than half of UTUCs [5, 17], and accounts for ≈25% of all renal carcinomas in China . Lastly, contrary to what has been observed in Western countries, female patients are less likely to develop UTUC tumours with high stage, large size, and LNM, as compared with men . In recent studies, there are no published reports describing the predictive factors for worse pathological outcome in the Chinese population. As the leading urological centre in China, we conducted the present study to identify predictive factors in this unique population.
Patients and Methods
We retrospectively collected the clinicopathological data of all consecutive patients who underwent RNU or distal ureterectomy at our centre from January 2002 to December 2010 (729 patients). After excluding distant metastasis, RNU or distal ureterectomy (in solitary kidney patients or patients with chronic kidney disease [CKD] grade IV–V with the evidence of ipsilateral functional kidney) was performed in patients with typical UTUC radiography or with positive cytology with or without fluorescence in situ hybridisation assay. Ureteroscopy with tumour biopsy was performed when radiography was atypical to determine treatment strategy. Patients with positive pathological evidence received surgical treatment and patients with negative evidence received re-ureteroscopy or were closely followed up. All the pathological data analysed in this study were identified from the surgical specimens. Estimated GFR (eGFR) was calculated with the re-expressed Modification of Diet in Renal Disease formulas for the Chinese population . Classification of CKD was based on the five-stage Kidney Disease Outcomes Quality Initiative (KDOQI) clinical practice guidelines .
All surgical specimens were processed according to standard pathological procedures. Tumour stage was assessed according to the Union for International Cancer Control (UICC) TNM classification of malignant tumours 2002. Tumour grade was assessed according to the WHO classification of 1973. Tumour architecture was defined as papillary or sessile. Tumour location was defined as pelvic/calyceal or ureteric. For tumours involving both sites, the location was attributed according to the most advanced stage and/or grade and/or the site of the dominant lesion. Tumour multifocality was defined as the simultaneous presence of two or more pathologically confirmed macroscopic tumours in any location(s) (renal pelvis or ureter). All urinary cytology specimens included both voided and catheterised bladder specimens, and bilateral selective ureteric washings. Urine cytology was classified as positive when malignant or atypical cells highly suggestive of UC were present. Cytology was classified as negative either in the absence of atypia, or, with atypia without mention of malignancy, or with scant cellular material.
Ipsilateral hydronephrosis was classified as present or absent, based on upper urinary tract imaging using either CT or MRI (with or without i.v. contrast), or, by IVU. Only imaging studies performed ≤6 weeks of RNU or distal ureterectomy were included for evaluation.
Univariate and multivariate logistic regression analysis was used to address the prediction for worse pathological outcomes (MID, NOC and Grade 3 histology). The chi-square test was used to test the distribution of categorical variables, and the Mann–Whitney U-test was used for continuous variables, e.g. age of onset.
Multivariable logistic regression coefficients were used to generate a nomogram for predicting worse pathological outcomes. To validate our regression model, we assessed discrimination and calibration. Discrimination was measured using the Harrell's concordance index (c-index), which is similar to the area under the receiver operating characteristic curve. Calibration was measured by calibration plots, which were generated to explore the nomograms performance using 200 bootstrap resamples. The generation of the nomogram and calibration plots was performed with the R open-source statistical software, and other statistical tests were performed with SPSS17.0 software. A P < 0.05 was considered to indicate statistical significance.
In all, 729 consecutive patients who had undergone RNU or distal ureterectomy at our centre during this period were identified. Of these 729 patients, 411(56.4%) were female, and the ratio of male to female was 1:1.29. There were 383 (52.7%) ureter TCCs in all the 727 patients with known tumour location. The mean (range) age of the 729 patients was 66.5 (20–94) years. Complete pathological and follow-up data was available for 693 patients. The distribution of clinicopathological characteristics are shown in Table 1.
Table 1. Clinical and pathological characteristics of the 693 patients treated with RNU or distal ureterectomy for UTUC
Primary tumour location:
CKD (eGFR, mL/min):
The results for predicting MID UTUCs are shown in Table 2. By multivariate analysis, male gender (hazard ratio [HR] 1.898, P = 0.001), sessile tumour architecture (HR 3.249, P < 0.001), high histological grade (HR 5.007, P < 0.001), preoperative ipsilateral hydronephrosis (HR 4.768, P < 0.001), tumour location within the renal pelvis or calyx (HR 2.620, P < 0.001), and tumour without multifocality (HR 1.639, P = 0.028) were significant predictive factors for MID UTUCs at the time of extirpation surgery. We used multivariable logistic regression coefficients to generate a corresponding nomogram (Fig. 1A) and calibration plot (Fig. 1B) for these data. The accuracy of the model nomogram measured by c-index was 0.786. Calibration plots revealed minimal underestimation at the ranges of low predicted probabilities for MID UTUCs.
Table 2. Predictive factors for MID UTUCs in 693 patients using univariate and multivariate analysis
*Analysis of the 498 patients with the known urine cytology results.
Predictive factors for NOC UTUCs in the 693 patients using univariate and multivariate analysis are shown in Table 3. By multivariate analysis, male gender (HR 2.132, P < 0.001), tumour location in renal pelvis or calyx (HR 3.466, P < 0.001), tumour without multifocality (HR 2.532, P = 0.001), sessile tumour architecture (HR 3.274, P < 0.001), and high histological grade (HR 3.019, P < 0.001) were each significantly predictive for NOC UTUC disease. Based on these multivariable logistic regression coefficients, we generated a corresponding nomogram (Fig. 2A) and calibration plot (Fig. 2B). The accuracy of the model nomogram measured by c-index was 0.792. Calibration plots revealed minimal under or overestimation at the range of predicted probabilities for NOC UTUC.
Table 3. Predictive factors for NOC UTUCs in 693 patients using univariate and multivariate analysis
*Analysis of the 498 patients with the known urine cytology results.
Predictive factors for Grade 3 histology in 498 patients using univariate and multivariate logistic regression are shown in Table 4. By multivariate analysis, chronological old age (HR 1.047, P < 0.001), sessile tumour architecture (HR 25.192, P < 0.001), preoperative ipsilateral hydronephrosis (HR 1.689, P = 0.024), and positive urinary cytology (HR 1.997, P = 0.006) were each predictive for histological Grade 3 UTUC.
Table 4. Predictive factors for Grade 3 histology in 498 patients with UTUC using univariate and multivariate logistic regression
0.966 (0.608–1.534 )
In the present study we found that male gender, sessile architecture, high histological grade, ipsilateral hydronephrosis, tumour location in the renal pelvis, tumour without multifocality and positive urine cytology results were predictive factors for worse pathological outcomes of UTUC in this Chinese population. Because the accurate prediction of worse pathological outcome could assist the urologist in clinical risk stratification and devising therapeutic options, we analysed every predictor in detail.
Of these factors, some are distinctive to the Chinese UTUC population. First, male gender as an independent predictive factor for MID and NOC was contrary to that in Western populations . Second, there was more female than male patients in this Chinese UTUC population (Yang et al.  from Taiwan reported that female patients had a significantly higher proportion of UTUC in all TCCs, and it was further confirmed by the present study with a male to female ratio of UTUC of 1 to 1.29.) Third, ureteric TCC was more common than renal pelvis TCC in this Chinese population. In the present study, primary tumour location in the ureter was found in 53.1% of all UTUCs. Lastly, we found that tumour without multifocality was significantly associated with MID and NOC in multivariate analysis. But the univariate analysis of Chromecki et al.  based on Western patients showed that UTUCs with multifocality had a higher rate of high stage, high grade, LNM, and sessile architecture, which was also contrary to the present results.
Given the discordance between the present findings and what others have reported, we further analysed gender-related differences (Table 5) and multifocality related differences (Table 6) regarding clinicopathological features. The present results show that the rate of severe CKD stage (P = 0.003) and high tumour stage (P < 0.001) was higher among women with UTUC. The rate of hydronephrosis (P = 0.037), papillary architecture (P = 0.002), CKD stage V (P = 0.005), and low tumour stage (P = 0.002) were all significantly higher in patients with tumour multifocality. By univariate logistic regression analysis, CKD stage V was still significantly associated with tumour multifocality (HR 1.206, 95%CI 1.001–1.452, P = 0.048). Both female patients and patients with tumour multifocality had a high rate of severe CKD. Use of Traditional Chinese Medicine (TCM) in China is common, and especially so among females. According to some existing theories, TCM containing aristolochic acid could contribute to the development of TCC  and CKD . Aristolochic acid-associated nephropathy has also been associated with multifocal disease in UC . We hypothesise that the use of TCM containing aristolochic acid may explain the observed incidence, gender differences and tumour location among Chinese with UTUC. Future studies should include a prospective epidemiological survey in the Chinese population to compare the difference of clinicopathological characteristics of UTUCs according to the exposure to aristolochic acid, which may further clarify this interesting phenomenon.
Table 5. Gender-related differences of the clinicopathological characteristics of UTUC
Pelvic TCCs were more common in Western countries, while ureteric TCC was more common in China. But results from China and Western countries show pelvic TCCs were more likely to develop as advanced disease than those in the ureter. Margulis et al.  reported that tumour location in the renal pelvis was significantly associated with NOC UTUCs. Isbarn et al.  reported that the proportion of tumours with high tumour stage and LNM was also higher in pelvis TCCs. In the present cohort, tumour location in the renal pelvis was an independent predictor for MID and NOC. To explain why TCC in the renal pelvis is associated with a higher rate of advanced disease, Raman et al.  suggested that the reason is that ureteric carcinomas, which quickly become symptomatic due to ureteric obstruction, are diagnosed earlier (and are thus detected at a lower stage). Earlier detection could even outweigh the fact that there is a thinner adventitial wall for the ureteric tumours to invade. To test the theory we further analysed the tumour location-related differences of clinicopathological characteristics in UTUCs, which are shown in Table 7. The results showed that presence of UTUC with Grade 3 histology (P = 0.003) and with sessile architecture (P < 0.001) were significantly higher in ureter TCC than pelvis TCC, which means that the aggressiveness of TCCs in the ureter were higher than those in the pelvis. However, the proportion of high tumour stage was significantly lower (P < 0.001) in ureteric TCC, while the proportion of ipsilateral hydronephrosis was significantly higher in ureteric TCC (P < 0.001). These results are consistent with Raman's hypothesis.
Table 7. Location-related differences of the clinicopathological characteristics of UTUC
We also noticed that there was almost no urethral UC in patients with TCC. Why this is remains unclear. We speculated that the cumulative contact time between the urothelial epithelium and carcinogenic materials in urine may be one important factor for the occurrence of UC. The cumulative contact time between urine and the urothelium would be longest within the bladder, owing to its function of urine storage, and shortest in the urethra. Also, the renal pelvis also stores urine, and would likely result in a longer contact time between urine and urothelium. These factors could potentially explain the difference in age of onset of UC in different locations and the rarity of urethral TCC.
Sessile architecture was associated with worse pathological outcome and poor prognosis. One large multicentre report  supported that sessile architecture was associated with high tumour stage, high histological grade, high rate of disease recurrence and cancer-specific mortality. In the present series, architecture was significantly associated with MID, NOC and Grade 3 diseases. The microscopic tumour architecture is available from routine ureteroscopic examination of the affected collecting system. Special attention should be paid on primary treatment and follow-up in UTUCs with sessile tumour architecture.
Recently, several articles have focused on the association between ipsilateral hydronephrosis and worse pathological outcome in UTUCs. Yet, the results are not consistent across series. Favaretto et al.  reported that ipsilateral hydronephrosis was not associated with MID and NOC disease in multivariate analysis. But Messer et al.  reported that hydronephrosis was an independent predictive factor for MID, NOC and high-grade UTUCs. In the present series, hydronephrosis was significantly associated with MID and grade 3 diseases, while there was no association between hydronephrosis and NOC disease. Information about hydronephrosis may improve the accuracy of pathological prediction and enable the selection of neoadjuvant chemotherapy or lymphadenectomy in appropriate patients.
The present study has certain limits and constraints. First, the research data represent a retrospective review of findings at a single centre. Second, patients who were not treated surgically were not included in the present analysis. Third, lack of information on follow-up and molecular biomarkers may reduce the strength of the present study. Nonetheless, further studies are necessary to confirm the role of molecular biomarkers as predictors for worse pathological outcome of UTUCs.
In conclusion, among a Chinese population, UTUC was more common in women than men. Primary UTUC was also found to be more common in the ureter than pelvis. Factors that predicted worse pathological outcomes among the present population included: male gender, sessile architecture, high histological grade, ipsilateral hydronephrosis, tumour location in the renal pelvis, tumour without multifocality and positive urine cytology results. Internal validation of the present analysis showed that the predictive nomogram for MID UTUC and NOC UTUC performed well.